Nowcasting earthquake damages with Twitter

Abstract

The Modified Mercalli intensity scale (Mercalli scale for short) is a qualitative measure used to express the perceived intensity of an earthquake in terms of damages. Accurate intensity reports are vital to estimate the type of emergency response required for a particular earthquake. In addition, Mercalli scale reports are needed to estimate the possible consequences of strong earthquakes in the future, based on the effects of previous events. Emergency offices and seismological agencies worldwide are in charge of producing Mercalli scale reports for each affected location after an earthquake. However, this task relies heavily on human observers in the affected locations, who are not always available or accurate. Consequently, Mercalli scale reports may take up to hours or even days to be published after an earthquake. We address this problem by proposing a method for early prediction of spatial Mercalli scale reports based on people's reactions to earthquakes in social networks. By tracking users' comments about real-time earthquakes, we create a collection of Mercalli scale point estimates at municipality (i.e., state subdivisions) level granularity. We introduce the concept of reinforced Mercalli support, which combines Mercalli scale point estimates with locally supported data (named 'local support'). We use this concept to provide Mercalli scale estimates for real-world events by providing smooth point estimates using a spatial smoother that incorporates the distribution of municipalities in each affected region. Our method is the first method based on social media that can provide spatial reports of damages in the Mercalli intensity scale. Experimental results show that our method is accurate and provides early spatial Mercalli reports 30 minutes after an earthquake. Furthermore, we show that our method performs well for earthquake spatial detection and maximum intensity prediction tasks. Our findings indicate that social media is a valuable source of spatial information for quickly estimating earthquake damages.

Using remarkability to define coastal flooding thresholds

“…We propose using the remarkability of a particular high-tide event, as measured by the volume of tweets about flooding generated in a particular day, as a measure of flood occurrence and severity. Other scholars have used social media data to identify damage 13,14 and aid management 15,16 of severe natural disasters, such as earthquakes [17][18][19] , heat waves 20 , hurricanes 21,22 , snowstorms 23 , and wildfires 24 . Researchers have also recently examined the ability to use social media to detect public attention paid to other climatic factors 25 .…”

Assessing the Intensity of the Population Affected by a Complex Natural Disaster Using Social Media Data

“…With the development of location-enabled mobile devices and the prosperity of social networks, social media provide an additional vital data source for emergency responses [15,29]. Much of the existing natural disaster research on using social media data have focused on various aspects, such as damage assessment [11,13,14,30] and event detection [31][32][33][34][35]. However, the application of social media data to improve the situation awareness of emergency responders is still in an early stage, and more efforts on this topic are needed [36].…”

“…In addition, social media data have proved to be capable of revealing the disaster-affected population [9]. First, data derived from social media have been shown to have strong relationships with the extent of damage from natural disasters [10][11][12][13]. Second, combined with other disaster-related data, such as geographical information and remote sensing images, social media data can enhance and improve useful information extracted for emergency response [14][15][16][17].…”

Exploitation of Social Network Data for Forecasting Garment Sales

“…In a similar study done by [12], AR sentiment aware model has been applied to forecast rev-enue on box office by extracting sentiments from online blogs and their influence on sales was found significant. Another domain, such as earthquake forecasting, where twitter data is used to predict damages caused by earthquakes is studied in [13]. In this study, machine learning models such as naïve Bayes (NB) and support vector machine (SVM) are used to classify tweets related to earthquake incidents and further embedded with spatial smoothing and regression models to estimate the loss due to earthquake.…”

“…Algorithm for assigning score to the word class.Sales data of fashion garment industry was aggregated weekly and an EF was performed. "Corrective Coefficient (V s )" was calculated, which represents the variation in sales as shown in Eq (13)…”